So far, I've just tried two transformers. One that puts out 11VAC at 1.2VA and another that puts out 13.5VAC at 12VA. The 11VAC unit causes all the bars on the ESM1 to light up but they never extinguish. Until I tried that I hadn't noticed that even the stock PS causes what appears to a be a boot-up sequence in which all the LEDs flash briefly.
The second transformer, a hefty beastie that once powered a heated massage cushion, produced results that were basically similar to the stock power supply. It's hard to make precise estimates because the bars tend to flicker a bit. The unloaded voltage for the stock PS is about 13.5VAC. I have a power supply that's marked 13.5VAC and actually reads about 16VAC, but I am reluctant to test it on something that provides a higher voltage than the stock transformer because it might cause the magic smoke to escape!
The next test will be to see if 100' of 16 gauge speaker wire will attenuate the signal, and if so, how much. If I can still get adequate signal with that much wire, I'm going to consider the multiple transformer phase of the testing complete, since you've established other transformers work as well. I'll then move on to selecting the components and creating the circuit design to read the ESM1 LEDs via phototransistors and sound an alarm if the meter head detects an X-10 type signal for more than 30 seconds. I don't care if it's noise or valid X-10 commands because anything that "blows a 10" on the ESM1 meter is capable of obliterating valid X-10 commands on the line.
I'd like to create a circuit that's able to differentiate between noise and valid X-10 signals, if only for troubleshooting the source of the broadcast storm. That should be pretty simple. All I need to do is to check whether the "X-10 Good" LED lights up along with the segments of the bar graph. I think it would be a good idea to monitor the very first segment of the LED bargraph, too, because lately I've been seeing the first segment light up and stay lit when the basement shoplites are running. I might decide to have the audio alert play a different tone if the background noise level goes beyond that first segment.
It looks as if the LEDs are close enough to the surface of the meter display window for me to be able to read them without spillover, but I won't know until I actually start trying to read the pulses whether that's going to be an issue. I think the best thing to use to butt up against the EMS1 display would be an 1/8" thick piece of black rubber with holes drilled in it to precisely line up with display LEDS. That should be easy to do. I'll take a piece of tracing paper and place it over the display head so I can mark the positions and then take that paper, tape it over the rubber strip that I'll tack to a piece of thin wood and then I'll drill holes through the paper, rubber and wood. The thin strip of wood will be thick enough to hold the phototransistors securely in place. I can bend the leads at a right angle and mount the circuit board horizontally in the plastic box that forms the mating part of the cradle.
FWIW, I tried to induce reading errors by moving the ESM1 head all around the wall wart. I had remarkably consistent readings. I did notice that the parallax error arising from the clear plastic lens and bargraph interface could easily lead someone to believe they were seeing one less bar when viewing the device from an angle. That parallax error means I may have to shroud the phototransistors carefully to insure they read just the LED they're aimed at. I hope it won't come to removing the clear plastic lens from the front of the meter to assure a tight optical coupling, but it could come to that.
Next thing to do is get some phototransistors and see what sort of output I can get from a flickering LED and then how to turn that output into a a multipoint audio alarm.
-- Bobby G.